Summary of the invention
The embodiment of the invention provides a kind of automatic tracing identification system and method, by using the variable field of view video camera, automatically follow the trail of the detail pictures of the characteristic portion of gathering suspicious object, can effectively realize the identification of suspicious object identity, carry out early stage, the reliable warning.
For achieving the above object, on the one hand, the embodiment of the invention provides a kind of automatic tracking system, comprising:
At least one is the visual field video camera fixedly, is used to gather the fixedly interior video image of guarded region scope, obtains first video image;
The suspicious object determination module is connected with described fixedly visual field video camera, is used for described first video image is analyzed, and determines suspicious object, and obtains the position of described suspicious object;
The trajectory track module is connected with described suspicious object determination module, is used for the position according to described suspicious object, obtains the movement locus of described suspicious object;
Control module is connected with described trajectory track module with described suspicious object determination module, is used for position and movement locus according to described suspicious object, the output control signal;
The variable field of view video camera is connected with described control module, is used for adjusting parameter based on described control signal, obtains meeting second video image of feature recognition requirement;
Picture recognition module, be connected with described variable field of view video camera with described control module, be used for described second video image is analyzed, determine the characteristic portion of described suspicious object, described characteristic portion is carried out feature identification, obtain the characteristic parameter of described suspicious object;
Comparing module is connected with described picture recognition module, and the characteristic information that is used for comparison object that the characteristic parameter and the property data base of described suspicious object are stored is compared;
Alarm module is connected with described comparing module, is used for reporting to the police when described comparison result indicates described suspicious object to meet the alert if of setting.
The multiobjective analysis module, be connected with described control module, be used for a plurality of suspicious object are carried out cluster analysis, judge whether these a plurality of suspicious object have association, when these a plurality of suspicious object have association, by described control module output control signal, control the described variable field of view camera acquisition characteristic portion image of suspicious object the most clearly, utilize described picture recognition module to carry out feature identification then; When these a plurality of suspicious object do not have association, then adopt the mode of poll, by described control module output control signal, control described variable field of view video camera and gather the characteristic portion detail pictures of suspicious object one by one, utilize described picture recognition module respectively a plurality of suspicious object to be carried out feature identification then.
Described fixedly visual field video camera is rifle formula video camera or ball-shaped camera;
Described variable field of view video camera is high speed monopod video camera or quick ball-shaped camera.
Described fixedly visual field video camera and described variable field of view video camera are ccd video camera, cmos camera or thermal imaging camera.
Described suspicious object determination module further comprises:
Extract submodule, be used for extracting the information of target travel object from described first video image;
The classification submodule is connected with described extraction submodule, is used for the information according to described target travel object, to described target travel sorting objects;
Judge submodule, be connected with described classification submodule, be used for classification, judge whether described target travel object violates predefined rule of conduct according to described target travel object, the target object of violating rule of conduct is defined as suspicious object, and obtains the position of described suspicious object.
Described trajectory track module further comprises:
Center of gravity is determined submodule, is used for the classification according to described suspicious object, determines the center of gravity of described suspicious object;
Track is determined submodule, determines that with described center of gravity submodule is connected, and is used for the position according to described suspicious object, writes down the track of described center of gravity, obtains the movement locus of described suspicious object.
Described picture recognition module further comprises:
Characteristic portion is determined submodule, is used for extracting from described second video image information of described suspicious object, determines its characteristic portion;
The recognin module determines that with described characteristic portion submodule is connected, and is used for described characteristic portion is carried out feature identification, obtains the characteristic parameter of described suspicious object.
Described picture recognition module also comprises:
Judge submodule, determine that with described characteristic portion submodule is connected with described recognin module, be used to judge whether described characteristic portion meets the requirement of feature identification, if meet the requirements, then the information with described characteristic portion sends to described recognin module, if undesirable, then notify described control module;
Described control module then is further used for according to feature recognition requirement, revises and the output control signal.
Also comprise:
The rule of conduct setting module, be connected with described picture recognition module, be used to set the decision condition of violating rule of conduct, the decision condition of described violation rule of conduct comprises: cross the border, enter the security area, the direction of motion is unusual, movement velocity is unusual, steal, leave over article and illegal parking.
Also comprise:
Database typing module is connected with described comparing module, is used for the characteristic information of described comparison object is entered into described property data base;
The alert if setting module is connected with described alarm module, is used for setting as required the condition of warning; The condition of described warning comprises: meet the warning of described comparison material body characteristics information and do not meet the warning of described comparison material body characteristics information.
Also comprise:
Matching module, be connected with described control module, be used to adjust the parameter of described variable field of view video camera, described variable field of view video camera and described fixedly visual field video camera are carried out the scene coupling, obtain the coarse adjustment angle of the described variable field of view video camera corresponding with the specific location point in the camera supervised regional extent in described fixedly visual field;
Described control module is further used for when described suspicious object is positioned at the specific location point of the camera supervised regional extent in described fixedly visual field, matching result according to described matching module, export control signal, control the coarse adjustment angle of the described specific location point correspondence of described variable field of view video camera quick rotation.
Also comprise:
The mapping output module is connected with described trajectory track module, is used for the movement locus coordinate of described suspicious object is mapped to the architectural plane coordinate at the current place of described suspicious object, and the architectural plane coordinate of described suspicious object is exported.
Also comprise:
The background light source module is used for when the ambient light of guarded region is lower than detection criterion, starts background light source;
Light monitoring and control module are connected with described background light source module, are used for guarded region is carried out the ambient light analysis, when the ambient light of guarded region is unfavorable for surveying, controls described background light source module and start.
On the other hand, the embodiment of the invention also provides a kind of automatic track identification method, may further comprise the steps:
To fixing visual field camera acquisition to first video image analyze, determine suspicious object, and obtain the position of described suspicious object;
According to the position of described suspicious object, obtain the movement locus of described suspicious object;
According to the position and the movement locus of described suspicious object, the output control signal;
Adjust the variable field of view camera parameters based on described control signal, obtain meeting second video image of feature recognition requirement;
Described second video image is analyzed, determined the characteristic portion of described suspicious object, described characteristic portion is carried out feature identification, obtain the characteristic parameter of described suspicious object;
The characteristic information of the comparison object stored in the characteristic parameter of described suspicious object and the property data base is compared;
When described comparison result indicates described suspicious object to meet the alert if of setting, report to the police.
Described method also comprises:
A plurality of suspicious object are carried out cluster analysis, judge whether these a plurality of suspicious object have association;
When these a plurality of suspicious object have association, export described control signal, control the described variable field of view camera acquisition characteristic portion image of suspicious object the most clearly, carry out feature identification then;
When these a plurality of suspicious object do not have association, then adopt the mode of poll, export described control signal, control the characteristic portion detail pictures that described variable field of view video camera is gathered suspicious object one by one, respectively a plurality of suspicious object are carried out feature identification then.
Described fixedly visual field video camera is rifle formula video camera or ball-shaped camera;
Described variable field of view video camera is high speed monopod video camera or quick ball-shaped camera.
Described fixedly visual field video camera and described variable field of view video camera are ccd video camera, cmos camera or thermal imaging camera.
Described to fixing visual field camera acquisition to first video image analyze, determine suspicious object, and the method for obtaining the position of described suspicious object is specially:
From described first video image, extract the information of target travel object;
According to the information of described target travel object, to described target travel sorting objects;
According to the classification of described mark moving object, judge whether described target travel object violates predefined rule of conduct, the target object of violating rule of conduct is defined as suspicious object, and obtains the position of described suspicious object.
Described position according to described suspicious object, the method that obtains the movement locus of described suspicious object is specially:
According to the classification of described suspicious object, determine the center of gravity of described suspicious object;
According to the position of described suspicious object, write down the track of described center of gravity, obtain the movement locus of described suspicious object.
Described described second video image is analyzed, determined the characteristic portion of described suspicious object, described characteristic portion is carried out feature identification, the method for characteristic parameters that obtains described suspicious object is specially:
From described second video image, extract the information of described suspicious object, determine its characteristic portion;
Described characteristic portion is carried out feature identification, obtain the characteristic parameter of described suspicious object.
Described described second video image is analyzed, determined the characteristic portion of described suspicious object, described characteristic portion is carried out feature identification, the method for characteristic parameters that obtains described suspicious object is specially:
From described second video image, extract the information of described suspicious object, determine its characteristic portion;
Judge whether described characteristic portion meets the requirement of feature identification;
If meet the requirements, then described characteristic portion is carried out feature identification identification, obtain the characteristic parameter of described suspicious object;
If undesirable, then revise and export control signal, control described variable field of view video camera and adjust parameter.
Also comprise:
Set to violate the step of the decision condition of rule of conduct, the decision condition of described violation rule of conduct comprise cross the border, enter the security area, the direction of motion is unusual, movement velocity is unusual, steal, leave over article and illegal parking.
Also comprise:
The feature ginseng information of described comparison object is entered in the described property data base, and the step of the condition of set reporting to the police as required, the condition of described warning comprises: meet the warning of described comparison material body characteristics information and do not meet the warning of described comparison material body characteristics information.
Also comprise:
Adjust the parameter of described variable field of view video camera, described variable field of view video camera and described fixedly visual field video camera are carried out the scene coupling, obtain the coarse adjustment angle of the described variable field of view video camera corresponding with the specific location point in the camera supervised regional extent in described fixedly visual field;
When described suspicious object is positioned at the specific location point of the camera supervised regional extent in described fixedly visual field,, control the coarse adjustment angle of the described specific location point correspondence of described variable field of view video camera quick rotation according to matching result output control signal.
Also comprise:
The movement locus coordinate of described suspicious object is mapped in the architectural plane coordinate at the current place of described suspicious object, and the step that the architectural plane coordinate of described suspicious object is exported.
Also comprise:
Guarded region is carried out the ambient light analysis, when ambient light is lower than detection criterion, start the step of background light source.
The embodiment of the invention has following beneficial effect:
1. system adjusts parameter by control variable field of view video camera, can farthest gather the detail pictures of suspicious object characteristic portion, thereby effectively realize the feature identification of suspicious object, carries out early stage, the reliable warning;
2. system determines the movement locus of suspicious object by the image that collects is carried out intellectual analysis, can carry out track identification to the suspicious object of motion.
Embodiment
Be illustrated in figure 1 as the functional module framework schematic diagram of the automatic tracing identification system of the embodiment of the invention, system comprises: fixedly visual field video camera 1, suspicious object determination module 2, trajectory track module 3, control module 4, variable field of view video camera 5, picture recognition module 6, comparing module 7 and alarm module 8.
Wherein:
Fixedly the visual field video camera 1, is used to gather the fixedly interior video image of guarded region scope, obtains first video image.
Suspicious object determination module 2 is connected with fixing visual field video camera 1, is used for first video image is analyzed, and determines suspicious object, and obtains the position of suspicious object.
Trajectory track module 3 is connected with suspicious object determination module 2, is used for the position according to suspicious object, obtains the movement locus of suspicious object.
Control module 4 is connected with trajectory track module 3 with suspicious object determination module 2, is used for position and movement locus according to suspicious object, the output control signal.
Variable field of view video camera 5 is connected with control module 4, is used for adjusting parameter based on the control signal of control module 4 outputs, obtains meeting second video image of feature recognition requirement.
Picture recognition module 6 is connected with variable field of view video camera 5 with control module 4, is used for second video image is analyzed, and determines the characteristic portion of suspicious object, and characteristic portion is carried out feature identification, obtains the characteristic parameter of suspicious object.
Comparing module 7 is connected with picture recognition module 6, and the characteristic information that is used for comparison object that the characteristic parameter and the property data base of suspicious object are stored is compared.
Alarm module 8 is connected with comparing module 7, is used for the result according to comparing module 7 comparisons, when comparison result indication suspicious object meets the alert if of setting, reports to the police.
Above-mentioned system can comprise one or more fixedly visual fields video camera 1, to obtain the video image of broad guarded region scope.Fixedly visual field video camera 1 can be rifle formula video camera or ball-shaped camera.According to the requirement of institute's environment for use, fixedly visual field video camera 1 video camera that can select common camera for use or have night vision function can be ccd video camera, cmos camera or thermal imaging camera.
Above-mentioned suspicious object determination module 2 further comprises:
Extract submodule, be used for extracting the information of target travel object from first video image.
The classification submodule is connected with the extraction submodule, is used for the information according to the target travel object, to the target travel sorting objects.
Judge submodule, be connected, be used for classification, judge whether the target travel object violates predefined rule of conduct, the target object of violating rule of conduct is defined as suspicious object, and obtains the position of suspicious object according to the target travel object with the classification submodule.
Suspicious object determination module 2 with the condition that the target travel object is defined as suspicious object is: whether this target travel object violates the rule of conduct of setting, and system at first also needs the decision condition of violating rule of conduct is set, and therefore, system also comprises:
The rule of conduct setting module is connected with suspicious object determination module 2, is used for setting the decision condition of violating rule of conduct according to the requirement of guarded region.
The decision condition of violating rule of conduct generally includes: enter the security area, cross the border, the direction of motion is unusual, movement velocity is unusual, steal, leave over article, illegal parking etc.
Determining the process of suspicious object, promptly is to judge whether the target travel object violates the process of the rule of conduct of setting, below violates the decision method of rule of conduct with specific embodiment explanation target travel object:
(1) judgement of crossing the border and invading
Set the warning line or the security area of guarded region at first as required, then the video image of gathering is analyzed, obtain the parameter such as movement position, the direction of motion of target travel object, when the movement position of target travel object exceeds or crossed over predefined warning line, or entered the security area, judge that promptly this target travel object violated rule of conduct, be defined as suspicious object.In some cases, need also to judge simultaneously whether the direction of motion of target travel object violates the rule of conduct of setting, when its direction of motion is also violated rule of conduct simultaneously, just be defined as suspicious object.
(2) movement velocity abnormality juding
Set the normal movement velocity scope of guarded region scope internal object moving object at first as required, can select the movement velocity of different directions, speed during for example horizontal direction is walked, the speed of falling etc., then the video image of gathering is analyzed, obtain parameters such as the movement velocity of target travel object and the direction of motion, when the target travel object exceeds the proper motion velocity interval of setting on the some direction of motion, this target travel object is defined as suspicious object.System also can carry out the movement velocity statistics at the colony that moves, and as judging that the foundation of violating rule of conduct is: when the speed of target travel object exceeds setting or adds up the group movement velocity interval that obtains, be defined as suspicious object.
Suspicious object determination module 2 can also at first carry out pre-alarm when determining suspicious object, note to remind the monitor staff.
Above-mentioned trajectory track module 3 further comprises:
Center of gravity is determined submodule, is used for the classification according to suspicious object, determines the center of gravity of suspicious object.
Track is determined submodule, determines that with center of gravity submodule is connected, and is used for the position according to suspicious object, writes down the track of suspicious object center of gravity, obtains the movement locus of suspicious object.
Trajectory track module 3 determines that the method for the movement locus of suspicious object is: the center of gravity of at first determining suspicious object according to the classification of suspicious object, as: when suspicious object is a man-hour, with people's centre of body weight or head as center of gravity, when suspicious object is vehicle, with the geometric center of vehicle as center of gravity, and, write down the track of its center of gravity according to the position of the suspicious object of determining in the suspicious object determination module 2, obtain the movement locus of suspicious object.
Control module 4 is obtained the position and the movement locus of the suspicious object that obtains in suspicious object determination module 2 and the trajectory track module 3, the output control signal, adjust parameter (as rotational angle, adjustment lens focus and aperture etc.) with control variable field of view video camera 5, can track the suspicious object of motion all the time with assurance variable field of view video camera 5, and collect the detail pictures of suspicious object.
Control module 4 can adopt the neural control method of class that variable field of view video camera 5 is carried out Based Intelligent Control.
Above-mentioned variable field of view video camera 5 can be high-velocity cloud platform video camera or quick ball-shaped camera, can capture the image of suspicious object in the full angle scope rapidly.Same, according to the requirement of institute's environment for use, the video camera that variable field of view video camera 5 also can be selected common camera for use or have night vision function can be ccd video camera, cmos camera or thermal imaging camera.
After variable field of view video camera 5 receives the control signal of control module 4, carry out parameter adjustment, the detail pictures of suspicious object is gathered, obtain second video image, and send picture recognition module 6 to.
Above-mentioned picture recognition module 6 further comprises:
Characteristic portion is determined submodule, is used for extracting from second video image information of suspicious object, determines its characteristic portion.
The recognin module determines that with characteristic portion submodule is connected, and is used for the characteristic portion of suspicious object is carried out feature identification, obtains the characteristic parameter of suspicious object.
When carrying out recognition of face, its characteristic parameter comprises: the colour of skin of people's face, interocular distance, skeletal structure, face feature etc., when carrying out car plate identification, its characteristic parameter comprises: the number-plate number, vehicle color etc.
In addition, in some cases, the image that variable field of view video camera 5 collects suspicious object might not meet picture recognition module 6 and carry out the requirement of feature identification, as the clear inadequately or characteristic portion information of characteristic portion information is imperfect etc. in the image that collects, at this moment, picture recognition module also comprises:
Judge submodule, determine that with characteristic portion submodule is connected with the recognin module, be used to judge whether the information of the characteristic portion of suspicious object meets the requirement of feature identification, if meet the requirements, then the information with characteristic portion sends to the recognin module, if undesirable, then notify control module 4.
Control module 4 is further used for the requirement according to feature identification, revises and the output control signal, adjusts parameters with control variable field of view video camera 5, the image of suspicious object characteristic portion is gathered, until the image of the requirement that obtains meeting feature identification.
For example, when characteristic portion information was clear inadequately in the image that collects, the judgement submodule then notified control module 4 characteristic portion information unintelligible, and control module 4 is revised and the output control signal, control variable field of view video camera 5 is adjusted focal length, with the size at amplification characteristic position; When characteristic portion is positioned at the image border so that characteristic portion information when imperfect, the judgement submodule then notifies control module 4 characteristic portion information imperfect, control module 4 is revised and the output control signal, control variable field of view video camera 5 rotational angles, make characteristic portion be in the middle position of image, to obtain complete characteristic portion image.
Preferably, after control module 4 control variable field of view video cameras 5 carry out parameter adjustment, make the characteristic portion (as people's face, license plate) of suspicious object in the video image collect be on the middle position and suitable scale of image, for example, make the characteristic portion of suspicious object account for entire image and be not less than on 15% the scale, be beneficial to picture recognition module 6 and carry out feature identification.
Form a closed-loop control system between picture recognition module 6, control module 4 and the variable field of view video camera 5, until second video image that obtains to meet feature recognition requirement.
Above-mentioned system also needs to finish the typing to characteristic information before utilizing comparing module 7 to carry out the characteristic parameter comparison, therefore, system also comprises:
Database typing module is connected with comparing module 7, is used for the characteristic information of comparison object is entered into property data base.
The characteristic information of described comparison object generally includes two classes: the first kind is: the clarification of objective information (for example fugitive suspect etc.) that needs detecting/warning; Second class is: special mandate and detect after the clarification of objective information (for example national treasury staff etc.) that will not report to the police.
The characteristic information that can adopt the mode of real-time typing will compare object is entered in the property data base, be specially: will need the comparison object of typing to carry out real-time data acquisition, and carry out the characteristic parameter that signature analysis obtains comparing object, then characteristic parameter is entered in the property data base; Also can adopt existing database is imported to mode in the property data base.
In addition, in order to guarantee the reliability of system, database typing module can be carried out real-time update to the information of preserving in the property data base.
Above-mentioned alarm module 8 can by start alarm lamp, sound or warning message is presented at the first-class multiple mode of monitor reports to the police gives the alarm, can also utilize relay that warning message is exported, for example, alarm module 8 can be connected with automatic door control system, when confirming that suspicious object is violated rule of conduct, by relay warning message is sent to automatic door control system, the control automatically-controlled door is closed, with the action of restriction suspicious object.
System also comprises: the alert if setting module is connected the alert if that is used for setting as required with alarm module 8.
The condition of reporting to the police generally includes: meet the warning of comparison material body characteristics information and do not meet the warning of comparison material body characteristics information.The warning that meets comparison material body characteristics information is promptly: when the characteristic information of suspicious object meets the clarification of objective information of above-mentioned first kind needs detecting/warning, report to the police; The warning that does not meet comparison material body characteristics information is promptly: during the clarification of objective information that will not report to the police after the characteristic information of suspicious object does not meet the special mandate of above-mentioned second class and detects, report to the police.
System can also comprise:
Memory module is used for after suspicious object is reported to the police, the video image of storage suspicious object.
In addition, when in above-mentioned suspicious object determination module 2 is judged the guarded region scope of fixing visual field video camera 1, occurring violating the suspicious object of rule of conduct, for make variable field of view video camera 5 can quick rotation to position that suspicious object occurs, feature detail pictures to suspicious object is gathered, and system can also comprise:
Matching module, be connected with control module 4, be used to adjust the parameter of variable field of view video camera 5, variable field of view video camera 5 and fixing visual field video camera 1 are carried out the scene coupling, obtain the coarse adjustment angle of the corresponding variable field of view video camera 5 of specific location point in the guarded region scope with fixing visual field video camera 1.
Variable field of view video camera 5 is specially with the fixing method of visual field video camera 1 scene coupling: at first by adjusting the angle and the focal length of variable field of view video camera 5, carry out one by one related with correspondence position point in fixing visual field video camera 1 monitor area specific location point in the variable field of view video camera 5 monitor area scopes, obtain the coarse adjustment angle of variable field of view video camera 5 corresponding each specific location point, its main purpose is: for variable field of view video camera 5 provides an initial rotation angle, guarantee variable field of view video camera 5 can quick rotation with fixing visual field video camera 1 monitor area scope in the coarse adjustment angle of specific location point coupling, and then the parameter by further fine setting variable field of view video camera 5 is with the detail pictures of collection suspicious object characteristic portion.
Certainly, along with moving of suspicious object, control module 4 also will be according to the movement locus of suspicious object, and control variable field of view video camera 5 is followed the tracks of and rotated and adjust focal length, the detail pictures of continuous acquisition suspicious object characteristic portion.
In order better to understand the particular location of suspicious object in current building, monitor the whereabouts of suspicious object, system also comprises:
The mapping output module is connected with trajectory track module 3, is used for the movement locus coordinate of suspicious object is mapped to the architectural plane coordinate at the current place of suspicious object, and the architectural plane coordinate of suspicious object is exported.
The mapping output module can be exported the architectural plane coordinate of suspicious object in monitor, is beneficial to the monitoring of monitor staff's suspicious object.
Simultaneously, consider that the ambient light in the guarded region may change along with the variation of available light, and then do not meet the requirement of image recognition that therefore, system also comprises:
Light monitoring and control module and are used for that guarded region is carried out the ambient light analysis, and when ambient light was unfavorable for surveying, control background light source module started.
The background light source module is connected with control module with the light monitoring, is used for when the ambient light of guarded region is lower than detection criterion, starts background light source.
The light monitoring can be with fixedly visual field video camera 1 and/or variable field of view video camera 5 be connected with control module, according to camera acquisition to image carry out the analysis of guarded region ambient light, certainly, also can directly adopt ambient light detection instrument etc., directly the ambient light of guarded region be surveyed.
In addition, when a plurality of suspicious object occurring in the guarded region scope, system also comprises:
Be used for multiobjective analysis module that a plurality of suspicious object that occur are simultaneously analyzed, be connected with control module 4, the multiobjective analysis module is at first carried out cluster analysis to a plurality of suspicious object, judge whether these a plurality of suspicious object have association, for example: whether the movement technique according to these a plurality of suspicious object is consistent, whether movement velocity is similar etc., and condition is judged, when these a plurality of suspicious object have association, by control module 4 output control signals, control variable field of view video camera 5 acquisition characteristics positions are the image of suspicious object the most clearly, utilizes picture recognition module 6 to carry out feature identification then; When the multiobjective analysis module analysis goes out a plurality of suspicious object and does not have association, then adopt the mode of poll, by control module 4 output control signals, control variable field of view video camera 5 is gathered the characteristic portion detail pictures of suspicious object one by one, utilizes picture recognition module 6 respectively a plurality of suspicious object to be carried out feature identification then.
System can also comprise the network equipment, be connected with memory module with alarm module 8, be used for communicating and being connected with other automatic tracing identification system or center host, when suspicious object is left the monitoring range of native system, the transmission of video images of warning message and suspicious object to other regional automatic tracing identification system or center host that network is connected, is prevented that suspicious object wanders away.
The network equipment can effectively enlarge the scope of guarded region, to be used to tackle security monitoring problem large-scale, important place.
In addition, system can also be connected with the safety monitoring system of outside other types, system comprises also and is used for the security monitoring transducer that is connected with the external security supervisory control system that outside safety monitoring system can be transferred the information that the variable field of view video camera be gathered to be needed by the security monitoring transducer.
The automatic tracing identification system that provides by above embodiment, adjust parameter by control variable field of view video camera, can farthest gather the detail pictures of suspicious object characteristic portion, thereby effectively realize the feature identification of suspicious object, carry out early stage, warning reliably; System determines the movement locus of suspicious object by the image that collects is carried out intellectual analysis, can carry out track identification to the suspicious object of motion.
Fig. 2 is the automatic track identification method flow schematic diagram of the embodiment of the invention, may further comprise the steps:
Step 21, to fixing visual field camera acquisition to first video image analyze, determine suspicious object, and obtain the position of suspicious object.
When determining suspicious object, can also at first carry out pre-alarm, note to remind the monitor staff.
Step 22 according to the position of suspicious object, obtains the movement locus of suspicious object.
Step 23, according to the position and the movement locus of suspicious object, the output control signal.
Step 24 is adjusted the variable field of view camera parameters based on control signal, obtains meeting second video image of feature identification.
Step 25 is analyzed second video image, determines the characteristic portion of suspicious object, and characteristic portion is carried out feature identification, obtains the characteristic parameter of suspicious object.
Step 26 is compared the characteristic information of the comparison object stored in the characteristic parameter of suspicious object and the property data base.
Step 27 when comparison result indication suspicious object meets the alert if of setting, is reported to the police.
The alert if of setting generally includes: meet the warning of comparison material body characteristics information and do not meet the warning of comparison material body characteristics information.
In the above-mentioned steps 21 to fixing visual field camera acquisition to first video image analyze, determine suspicious object, and the method for obtaining the position of suspicious object is specially:
From first video image, extract the information of target travel object;
According to the information of target travel object, to the target travel sorting objects;
According to the classification of target travel object, judge whether the target travel object violates predefined rule of conduct, the target object of violating rule of conduct is defined as suspicious object, and obtains the position of suspicious object.
The decision condition of violating rule of conduct generally includes: cross the border, the direction of motion is unusual, movement velocity is unusual, steal, leave over article and illegal parking etc., in case the target travel object is violated the rule of conduct of setting, promptly can be denoted as suspicious object.
Step 21 also comprises:
Set the step of the decision condition of violating rule of conduct.
According to the position of suspicious object, the method that obtains the movement locus of suspicious object is specially in the above-mentioned steps 22:
According to the classification of suspicious object, determine the center of gravity of suspicious object;
According to the position of suspicious object, write down the track of center of gravity, obtain the movement locus of suspicious object.
In the above-mentioned steps 25 second video image is analyzed, determined the characteristic portion of suspicious object, characteristic portion is carried out feature identification, the method for characteristic parameters that obtains suspicious object is specially:
From second video image, extract the information of suspicious object, determine its characteristic portion;
Characteristic portion is carried out feature identification, obtain the characteristic parameter of suspicious object.
In addition, in some cases, the variable field of view camera acquisition is to the image of suspicious object and do not meet the requirement of signature analysis, as the clear inadequately or characteristic portion information of characteristic portion information is imperfect etc. in the image that collects, therefore, in the above-mentioned steps 25 second video image is analyzed, determined the characteristic portion of suspicious object, characteristic portion is carried out feature identification, obtain suspicious object method of characteristic parameters can also for:
From second video image, extract the information of suspicious object, determine its characteristic portion;
Whether the judging characteristic position meets the requirement of feature identification;
If meet the requirements, then characteristic portion is carried out feature identification identification, obtain the characteristic parameter of suspicious object;
If undesirable, then revise and export control signal, control variable field of view video camera is adjusted parameter, the image of suspicious object characteristic portion is gathered, until the image that obtains meeting feature recognition requirement.
Preferably, after control variable field of view video camera is adjusted parameter (rotational angle and adjustment focal length, aperture etc.), the characteristic portion of suspicious object in the video image that collects (as people's face) is on the middle position and suitable scale of image, for example, making the characteristic portion of suspicious object account for entire image is not less than on 15% the scale.
Above-mentioned method also comprises:
The characteristic parameter of comparison object is entered in the property data base, and sets the step of alert if.
In addition, when in judging the guarded region scope of fixing visual field video camera, occurring violating the suspicious object of rule of conduct, for make the variable field of view video camera can quick rotation to position that suspicious object occurs, feature detail pictures to suspicious object is gathered, and above-mentioned method can also comprise:
To the variable field of view video camera and fixedly the visual field video camera carry out scene coupling, obtain the step of the coarse adjustment angle of the corresponding variable field of view video camera of specific location point in the guarded region scope with fixing visual field video camera; And
When suspicious object was positioned at the specific location point of fixing visual field video camera, according to matching result, control variable field of view video camera quick rotation was to the step of the coarse adjustment angle of specific location point correspondence.
In addition, in order better to understand the position of suspicious object in current building, monitor the whereabouts of suspicious object, said method also comprises:
The movement locus coordinate of suspicious object is mapped in the architectural plane coordinate at the current place of suspicious object, and the step that the architectural plane coordinate of suspicious object is exported.
And, consider that the ambient light in the guarded region may change along with the variation of available light, and then do not meet the requirement of image recognition that therefore, said method also comprises:
Guarded region is carried out the ambient light analysis, when ambient light is lower than detection criterion, start the step of background light source.
In addition, when a plurality of suspicious object occurring in the guarded region scope, at first a plurality of suspicious object are carried out cluster analysis, judge whether these a plurality of suspicious object have association, for example: conditions such as whether, movement velocity whether consistent according to the movement technique of these a plurality of suspicious object be similar are judged, when these a plurality of suspicious object had association, signature analysis was carried out at acquisition characteristics position suspicious object the most clearly; When a plurality of suspicious object do not have association, then adopt the mode of poll, gather the characteristic portion detail pictures of suspicious object one by one, respectively a plurality of suspicious object are carried out feature identification.
In addition, after system reports to the police to suspicious object, also need to store the video image of suspicious object, and the video image of warning message and suspicious object is transferred to other regional automatic tracing identification system that network is connected or center host etc. by the network equipment.
The automatic track identification method that provides by above embodiment, adjust parameter by control variable field of view video camera, can farthest gather the detail pictures of suspicious object characteristic portion, thereby effectively realize the feature identification of suspicious object, carry out early stage, warning reliably; System determines the movement locus of suspicious object by the image that collects is carried out intellectual analysis, can carry out track identification to the suspicious object of motion.
Fig. 3 is a concrete scene schematic diagram of the automatic tracing identification system of the embodiment of the invention, this automatic tracing identification system can be based on outer computer or DSP treatment system, and fixedly visual field rifle formula video camera, variable field of view video camera, track identification main frame, input-output equipment (monitor and keyboard) and the network equipment are formed by three in system.Wherein, fixedly visual field rifle formula video camera is gathered the video image in the guarded region scope; The track identification main frame is the integrated of a series of modules such as above-mentioned suspicious object determination module, trajectory track module, control module, picture recognition module, comparing module, alarm module, finishing the function of the target travel object being carried out automatic track identification, is the core of system; The variable field of view video camera is accepted to adjust parameter after the control signal of track identification main frame, and the characteristic portion detail pictures of suspicious object is gathered; System can be by monitor with the output of the video image of warning message and suspicious object, finishes operations such as typing to property data base by keyboard; System is network enabled device also, be used for being connected with other automatic tracing identification system, when suspicious object is left the monitoring range of native system, native system can be with information such as the data of suspicious object and the directions of motion by LAN or Internet network delivery to other moving track identification systems, prevent that suspicious object wanders away, the network equipment can effectively enlarge the scope of guarded region.
The above only is a preferred implementation of the present invention; should be pointed out that for those skilled in the art, under the prerequisite that does not break away from the principle of the invention; can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.